In this work, we use the Clouds and the Earth's Radiant Energy System (CERES) FluxByCloudTyp data product, which calculates TOA shortwave and longwave fluxes for cloud categories defined by cloud optical depth (τ) and cloud top pressure (pc ), to evaluate the HadGEM2-A model with a simulator. The CERES Flux-by-cloud type simulator is comprised of a cloud generator that produces subcolumns with profiles of binary cloud fraction, a cloud property simulator that determines the (τ, pc ) cloud type for each subcolumn, and a radiative transfer model that calculates TOA fluxes. The identification of duplicate atmospheric profiles reduces the number of radiative transfer calculations required by approximately 97.6%. In the Southern Great Plains region in JFD (January, February, and December) 2008, the simulator shows that simulated cloud tops are higher in altitude than observed, but also have higher values of OLR than observed, leading to a compensating error that results in an average value of OLR that is close to observed. When the simulator is applied to the Southeast Pacific stratocumulus region in JJA 2008, the simulated cloud tops are primarily low in altitude; however, the clouds tend to be less numerous, and have higher optical depths than are observed. In addition to the increase in albedo that comes from having too many clouds with higher optical depth, the HadGEM2-A albedo is higher than observed for those cloud types that occur most frequently. The simulator is also applied to the entire 60° N - 60° S region, and it is found that there are fewer clouds than observed for most cloud types, but there are also higher albedos for most cloud types, which represents a compensating error in terms of the shortwave radiative budget.